Contents lists available at ScienceDirect

Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat

Fate of antibiotic resistance genes in reclaimed water reuse system withintegrated membrane process

Jian Lua,b,c,⁎, Yuxuan Zhanga,b, Jun Wud, Jianhua Wanga, Ying Caia

a CAS Key Laboratory of Coastal Environmental Processes and Ecological Remediation, Yantai Institute of Coastal Zone Research (YIC), Chinese Academy of Sciences(CAS), Shandong Key Laboratory of Coastal Environmental Processes, YICCAS, Yantai, Shandong 264003, PR ChinabUniversity of Chinese Academy of Sciences, Beijing 100049, PR Chinac Center for Ocean Mega-Science, Chinese Academy of Sciences, 7 Nanhai Road, Qingdao, 266071, PR Chinad School of Resources and Environmental Engineering, Ludong University, Yantai, Shandong 264025, PR China

G R A P H I C A L A B S T R A C T

A R T I C L E I N F O

Editor: Daniel C.W. Tsang

Keywords:Antibiotic resistance genesIntegrated membrane processReclaimed water reuse systemRemoval efficiencyBacterial community

A B S T R A C T

The fate of antibiotic resistance genes (ARGs) in reclaimed water reuse system with integrated membraneprocess (IMR) was firstly investigated. Results indicated that ARGs, class 1 integrons (intI1) and 16S rRNA genecould be reduced efficiently in the IMR system. The absolute abundance of all detected ARGs in the reuse waterafter reverse osmosis (RO) filtration of the IMR system was 4.03×104 copies/mL, which was about 2–3 ordersof magnitude lower than that in the raw influent of the wastewater treatment plants (WWTPs). Maximum re-moval efficiency of the detected genes was up to 3.8 log removal values. Daily flux of the summation of allselected ARGs in the IMR system decreased sharply to (1.02 ± 1.37) ×1014 copies/day, which was 1–3 ordersof magnitude lower than that in the activated sludge system (CAS) system. The strong clustering based onordination analysis separated the reuse water from other water samples in the WWTPs. Network analysis re-vealed the existence of potential multi-antibiotic resistant bacteria. The potential multi-antibiotic resistantbacteria, including Clostridium and Defluviicoccus, could be removed effectively by microfiltration and RO fil-tration. These findings suggested that the IMR system was efficient to remove ARGs and potential multi-anti-biotic resistant bacteria in the wastewater reclamation system.

https://doi.org/10.1016/j.jhazmat.2019.121025Received 27 March 2019; Received in revised form 14 August 2019; Accepted 14 August 2019

⁎ Corresponding author at: CAS Key Laboratory of Coastal Environmental Processes and Ecological Remediation, Yantai Institute of Coastal Zone Research (YIC),Chinese Academy of Sciences (CAS), Shandong Key Laboratory of Coastal Environmental Processes, YICCAS, Yantai, Shandong 264003, PR China.

E-mail addresses: jlu@yic.ac.cn, lujianleo@163.com (J. Lu).

Journal of Hazardous Materials 382 (2020) 121025

Available online 15 August 20190304-3894/ © 2019 Elsevier B.V. All rights reserved.

T

1. Introduction

Extensive and improper use of antibiotics in hospital, livestockfarming and agriculture-industrial production has led to the rapidprevalence of antibiotic resistant bacteria (ARB) and antibiotic re-sistance genes (ARGs), even the widespread occurrence of multi-anti-biotic resistant bacteria (MARB) and genes in various environments(Zhu et al., 2017; Zhou et al., 2011; Zhang et al., 2015; Su et al., 2017).ARGs might persist and disseminate widely by horizontal transfer me-chanism in the environment even in the absence of environmental se-lection pressure. Numerous studies have reported the occurrence, dis-tribution and removal of ARGs in different environments (Lu et al.,2019). The increasing emergence and propagation of antimicrobialresistance have been posing risks to ecological sustainability andhuman health (Berendonk et al., 2015).

Many developing countries are facing the challenge of watershortages. Reclaimed water has been identified as a feasible and ef-fective solution to water shortages (Wang et al., 2014). However, thelack of water quality assurance remains one of the most serious hurdlesfor the wide use of reclaimed water. As emerging contaminants, ARGshave become one of the important factors effecting water quality (Wanget al., 2014; Hu et al., 2018). Reclaimed water derived from wastewatertreatment plants (WWTPs) can relieve the serious challenge to someextent. WWTPs, the important receiver of domestic wastewater, med-ical wastewater, pharmaceutical wastewater and husbandry waste-water, have played critical roles in removing ARGs to alter the dis-tribution and abundance of ARGs in receiving environment (Tahraniet al., 2015; Proia et al., 2018). WWTPs not only serve as importantreservoirs of persistent ARGs and ARB from different sources, but alsoplay vital roles in reducing and controlling the proliferation and spreadof ARB and ARGs (Pruden et al., 2013). However, the conventionalactivated sludge (CAS) system in WWTPs cannot completely removeARGs and ARB, and it was influenced by different treatment processes.Therefore, the advanced treatment after conventional biological processand disinfection should be applied for improving the removal efficiencyof emerging contaminants (Michael et al., 2013), including reverseosmosis membrane filtration, activated oxidation processes, activatedcarbon adsorption, chlorination, ultraviolet and ozone disinfection, etc(Zheng et al., 2017). Many techniques such as coagulation (Li et al.,2017), anaerobic treatment (Yi et al., 2017; Ma et al., 2018), ultravioletdisinfection (Hu et al., 2016; Zhang et al., 2019), advanced chemicaloxidation (Michael-Kordatou et al., 2018; Fiorentino et al., 2019), andmembrane bioreactor system (Le et al., 2018) have been used on theelimination of ARGs. Zhang et al. (2016) discovered that Fenton oxi-dation process (Fe2+/H2O2) was effective on the reduction of ARGs.

However, cost of this technique was high due to consumption of largeamounts of reagents and it also brought secondary pollutants. Severalstudies found that the removal of ARGs in membrane bioreactor (MBR)was better than that in traditional processes (Le et al., 2018). However,systematic analysis and comparisons for the removal of ARGs by thesemethods have not been reported.

Membrane treatment is one of the promising solutions for removalof emerging contaminants. Theoretically, ARGs as high molecularcompounds could be trapped and removed by advanced membrane. Lanet al. (2019) found that nanofiltration and reverse osmosis filtrationprocesses were capable of removing various types of ARGs efficiently.However, limited studies are available on investigating the removal ofARGs and ARB in reclaimed water reuse system with integrated mem-brane process (IMR) consisting of conventional anaerobic-anoxic-aerobic (AAO) process, microfiltration, ultrafiltration, and reverse os-mosis treatment. Therefore, the objective of this study was to evaluatethe distribution and removal efficiency of ARGs and class 1 integron(intI1) in IMR system, microfiltration membrane bioreactor (MBR)system and conventional activated sludge system (CAS) in WWTPs.Previous studies demonstrated that antibiotics and corresponding genesencoding resistance to sulfonamides, tetracyclines, macrolides andquinolones have been detected frequently in the coastal zone of China.Therefore, tetracycline resistance genes (tetB, tetG and tetX), sulfona-mide resistance genes (sul1 and sul2), macrolide resistance genes (ermFand ermT), quinolone resistance genes (qnrA, qnrB and qnrS), intI1 and16S rRNA gene were investigated and quantified by real-time PCR.

2. Materials and methods

2.1. Sample collection

A typical WWTPs combining the CAS system with the IMR system inYantai City, was chosen to investigate the removal of ARGs. Besides, theeffluent of the WWTPs and seawater collected from the receiving seawere monitored for investigating the influence of the WWTPs on re-ceiving environment. Six wastewater samples (W1-1, W1-2, W1-3, W1-4, W1-5 and W1-6) and one sludge sample (W1-S1) were collected fromthe CAS system, eight water samples (W2-1, W2-2, W2-3, W2-4, W2-5,W2-6, W2-7 and W2-8) and four sludge samples (W2-S1, W2-S2, W2-S3and W2-S4) were collected from the IMR system, and three seawatersamples (WE1, WE2 and WE3) were collected from the receiving sea(Fig. 1). Samples were collected in sextuplicate from each samplingpoint, and two of them were mixed well as one subsample. Subsample-1, subsample-2 and subsample-3 were parallel samples in each sam-pling site. Subsequent DNA extraction, PCR amplification and statistical

Fig. 1. Schematic diagram of the reclaimed water reuse system with integrated membrane process (IMR) and conventional activated sludge system (CAS) in WWTPs.Upper-right corner, the map of seawater sampling locations.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

2

analysis were performed base on the three parallel samples, while thehigh-throughput sequencing was conducted based on the mixture ofsubsample-1, subsample-2 and subsample-3. WE1, WE2 and WE3 werecollected from the receiving sea. WE1 was located at the drainage outletof the WWTPs, WE2 was at the edge of pollutant discharge region of theWWTPs, and WE3 was about 2.0 km away from the outlet of WWTPs.Sludge samples were collected from aerobic tank (W1-S1) of the CASsystem, anaerobic tank (W2-S1), anoxic tank (W2-S2), aerobic tank(W2-S3), and microfiltration tank (W2-S4) of the IMR system. All watersamples and sludge samples were stored in sterile amber containerswith iceboxes and transported immediately to laboratory for furtheranalyses within the same day.

2.2. Physicochemical analysis

Salinity was measured using a portable refractometer (LH-Y100,Lohand Biological, China) and pH was determined by a pH meter (PHS-3C, INESA, China). Ammonia, nitrates, nitrites, phosphates, silicate,total nitrogen (TN) and total phosphorus (TP) were analyzed by acontinuous flow analyzer (Auto Analyzer III, Seal, Germany). Totalorganic carbon (TOC) was measured with a total organic carbon ana-lyzer (TOC-VCPH, Shimadu, Japan). Prior to analyses, water sampleswere filtered through 0.45 μm mixed cellulose esters membranes(Merck Millipore Ltd, Ireland). The water quality parameters wereshown in Table S1.

2.3. DNA extraction

Water sample with the volume of 1.0 L collected from each samplingpoint was filtered through 0.22 μm mixed cellulose esters membranes(Millipore) for DNA extraction according to previous studies (Lu et al.,2019; Hu et al., 2018), and 0.5 g sludge from each sample was used forDNA extraction. Extraction of DNA was performed by TIANamp SoilDNA Kit (TIANGEN Biotech, Beijing, China) according to manu-facturer’s instructions. Concentration and purity of DNA were measuredby a NanoDrop UV–vis spectrophotometer (NanoDrop Lite, ThermoScientific, Wilmington, USA), and the DNA quality was determined by1% agarose gel electrophoresis.

2.4. Quantification of ARGs

A total of twelve genes were quantified by real-time PCR (qPCR),including 16S rRNA gene, intI1 and ten types of ARGs. All target ARGsconferred resistance to common antibiotic families used in veterinaryand human medicine including sulfonamide resistance genes (sul1,sul2), tetracycline resistance genes (tetB, tetG, tetX), macrolide re-sistance genes (ermF, ermT) and quinolone resistance genes (qnrA, qnrB,qnrS). These target ARGs were detected with higher detection fre-quencies and higher abundances in our previous studies. All primersequences used in this study were summarized in Table S2, which werereported in previous publications (Zhu et al., 2013).

Calibration standard curves for absolute quantification and positivecontrols were established by a ten-fold dilution series of plasmids DNAranging from 107 to 100 copies/μL, which were extracted from E. coliDH5α as reported previously (Luo et al., 2010). All the gene standardswere run together with samples in 384-well plates and each reactionwas conducted in triplicate. Each 10 μL real-time PCR reaction con-tained 5.0 μL SYBR Premix Ex Taq II (Takara, Dalian, China), 0.4 μLforward primers and 0.4 μL reverse primers (0.4 μM), and 1.0 μL DNAtemplate. Real-time PCR amplified reactions were conducted using Bio-Rad qPCR system (Bio-Rad CFX384 Touch, CA, USA) with the followingprogram: 30 s for denaturation at 95 °C, 40 cycles of 5 s at 95 °C, 30 s atannealing temperature and 40 s for extension at 72 °C. Finally, amelting curve was conducted with temperature ranging from 60 °C to95 °C and the melting curves with a single peak were considered asspecific amplification. Reactions with amplification efficiencies ranged

from 90% to 110% and the R2 values of standard curves more than 0.99were taken into account for ARGs quantification.

2.5. Removal efficiencies of 16S rRNA gene, intI1 and ARGs

Removal efficiencies of 16S rRNA gene, intI1 and ARGs for the CASsystem and the IMR system were calculated in equation as follows:

=

−

−

Removal efficiency of CAS system lg cc

w1 1

w1 6 (1)

=

−

−

Removal efficiency of MF system lg cc

w2 1

w2 7 (2)

=

−

−

Removal efficiency of IMR system lg cc

w2 1

w2 8 (3)

The removal efficiencies of detected genes were assessed usinglog10 removal value (LRV) as depicted in Eqs. (1)–(3). The removalefficiency of the CAS system was calculated by Eq. (1). Eq. (2) de-monstrated the removal efficiency of the wastewater treatment flowfrom the raw influent to the effluent of microfiltration in the IMRsystem. The total removal efficiency of the IMR system was depicted byEq. (3), ranging from the raw influent to the reuse water.

2.6. Illumina MiSeq sequencing and analyses of 16S rRNA gene

Purified microbial DNA extracted from 17 water samples and5 sludge samples were sent to Majorbio (Shanghai, China) for high-throughput sequencing. The V4-V5 region of the bacterial 16S rRNAgene was selected for amplification with primers 515 F (5′-GTGCCAGCMGCCGCGG-3′) and 907R (5′−CCGTCAATTCMTTTRA GTTT -3′)(Wang et al., 2018), using thermocycler PCR system (GeneAmp 9700,ABI, USA). PCR reactions were conducted using the following program:3min of denaturation at 95 °C, 27 cycles of 30 s at 95 °C, 30 s for an-nealing at 55 °C, and 45 s for elongation at 72 °C, and a final extensionat 72 °C for 10min. PCR reactions were performed in triplicate. Each20 μL mixture containing 4 μL of 5 × FastPfu Buffer, 2 μL of 2.5 mMdNTPs, 0.8 μL of each primer (5 μM), 0.4 μL of FastPfu Polymerase and10 ng of template DNA. The PCR products were extracted from 2%agarose gels and further purified using the AxyPrep DNA Gel ExtractionKit (Axygen Biosciences, Union City, CA, USA). Purified amplicons werepooled in equimolar and paired-end sequenced (2×300) on an Illu-mina MiSeq platform (Illumina, San Diego, USA) according to thestandard protocols by Majorbio (Shanghai, China). A total of 989,644high quality 16S rRNA gene sequences were generated from the 22samples. After subsampling to an equal sequence, 2084 operationaltaxonomic units (OTUs) at 97% identity were obtained. Raw reads weredeposited into the National Center for Biotechnology Information(NCBI) Sequence Read Archive (SRA) database (Accession Number:SRR8490435 - SRR8490453, SRR8504322 - SRR8504324).

2.7. Statistical analysis

Statistical analysis and assessments were performed using Origin2019 (Origin Lab Corporation, USA) and SPSS 19 (IBM, USA). All sta-tistical tests were considered significant at p < 0.05. Canonical cor-respondence analysis (CCA) and non-metric multi-dimensional scaling(NMDS) analysis between water quality parameters, antibiotic re-sistance genes and bacterial communities were performed in R en-vironment using vegan and ggplot2 packages. Network analysis wasperformed in R using vegan, igraph and hmisc packages (Li et al., 2015)and further visualized by Cytoscape 3.7.1 to demonstrate the correla-tion between detected genes and bacterial communities.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

3

3. Results and discussion

3.1. Fate of ARGs in the effluent of the WWTPs and the receivingenvironment

In order to investigate the influence of WWTPs discharge on itsreceiving environment, the effluent of CAS system (W1-6), the effluentof microfiltration in IMR system (W1-7), and three seawater samplescollected from the surrounding sea (WE1, WE2, WE3) were chosen fordetection and quantification of target ARGs (sul1, sul2, tetB, tetG, tetX,ermF, ermT, qnrA, qnrB, qnrS), intI1 and 16S rRNA gene. All 12 targetgenes were detected in these five samples, except that qnrS was de-tected in the effluent of microfiltration (W2-7) with low abundance of0.46 copies/mL. The sul genes (sul1 and sul2), encoding sulfonamideresistant dihydropteroate synthase (Zhang et al., 2014), were the mostprevalent ARGs in these water samples with abundances ranging from2.14×103 to 2.25×106 copies/mL. The average abundances of tetgenes (tetB, tetG and tetX), which encoded tetracycline resistance pro-tein, were lower than those of sul genes but higher than those of qnr anderm genes (Tukey HSD Post-hoc Test, p < 0.05). The qnr genes (qnrA,qnrB and qnrS), encoding resistant to quinolones (Marti et al., 2014),were detected with abundances ranging from 46.0 to 2.48× 105 co-pies/mL. The abundances of qnrA were higher than those of qnrB andqnrS (p < 0.05). The abundances of erm genes (ermF and ermT) werethe lowest, which associated with those of methylase conferring mac-rolides resistance.

As shown in Fig. 2, the absolute copy numbers of sul1, sul2, tetX andqnrA were relatively high, ranging from 103 to 108 copies/mL. ExceptermT and qnrB genes, the absolute abundances of other ARGs in theeffluent of microfiltration (W2-7) were significantly lower than those inthe effluent of CAS system (W1-6) (p < 0.05). The absolute abun-dances of most ARGs (sul, tet, qnrA and qnrS) detected in seawater(WE1, WE2, WE3) were significantly higher than those in the effluent ofmicrofiltration (W2-7), while they were lower than the abundances ofgenes detected in the effluent of CAS system (W1-6) (p < 0.05).

The gene of IntI1 was important transmissible vector that playedimportant role in the emergence and horizontal transfer of ARGs indifferent environment (Xu et al., 2017). The intI1 gene was prevalent inthe water samples in this research. The absolute abundances of intI1detected in the receiving seawater ranged from 1.48× 105 to3.36×107 copies/mL, higher than that in the effluent of microfiltra-tion process (1.13×104 copies/mL) (p < 0.05) and comparable withthat in the discharge of the CAS system (8.87× 106 copies/mL)(p > 0.05). In addition, the abundances of 16S rRNA gene,

representing the microbial concentrations (Chen and Zhang, 2013),were obviously higher in the effluent of CAS system than those in theeffluent of microfiltration system (p < 0.05). Besides, the abundancesof 16S rRNA gene in seawater samples ranged from 103 to 5.0× 106

copies/mL, showing lower abundances than those in the effluent of CASsystem (p < 0.05).

It has been reported in our previous studies that the abundances ofARGs detected in coastal areas were obviously higher than those in theseawater (Lu et al., 2019). According to the investigation, the abun-dances of ARGs detected in W1-6 and WE (1, 2, 3) were 3–6 orders ofmagnitude higher than that in the seawater, and 1–3 orders of magni-tude higher than that in the coastal water. It could be speculated thatthe area near the drainage outlet of the WWTPs was influenced by theeffluent of WWTPs. The discharges of the WWTPs might contribute tothe spread of intI1 and ARGs into surrounding environment. Mean-while, in comparison with the CAS system, the abundance of ARGs andintI1 decreased obviously after the membrane filtration in the IMR

Fig. 2. Absolute abundances (copies/mL) of 16S rRNA,intI1 and ARGs. W1-6, the effluent of conventionalactivated sludge (CAS) system; W2-7, the effluent aftermicrofiltration in the reclaimed water reuse systemwith integrated membrane process (IMR); WE1, WE2and WE3, seawater samples collected from three sam-pling points located in the receiving sea, approxi-mately 0 km, 0.5 km and 2 km away from the WWTPsdischarge point.

Fig. 3. Heat map of absolute abundances of ARGs, 16S rRNA and intI1 in watersamples and sludge samples. Copies/mL, unit of water samples; copies/g, unitof sludge samples.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

4

system.

3.2. Removal of ARGs and intI1 in the IMR system

To further investigate the removal of ARGs and intI1 in the IMRsystem which included anaerobic-anoxic-aerobic (AAO) tank, micro-filtration (MF) process and RO filtration process, the absolute abun-dances of 16S rRNA gene, intI1 and four types of ARGs in this systemwere quantified. CAS system including AAO tank, secondary clarifier(SC) tank and ultraviolet (UV) disinfection process was used as thecomparison. The absolute abundances of detected genes were

demonstrated in the heat map (Fig. 3) and bar chart (Fig. 4).In the raw influent (W1-1 and W2-1) of the WWTPs, all target genes

were detected at relatively high abundances. The absolute abundancesof ARGs ranged from 3.46× 102 (tetB) to 2.58×106 (sul2) copies/mL.Sulfonamide resistance genes (sul1 and sul2) were the most prevalentARGs in the raw influent, followed by tetracycline resistance gene (tetX)and quinolone resistance gene (qnrA). IntI1 gene was detected at thehighest abundance up to 7.91× 107 copies/mL. As shown in Figs. 3 and4, there was no significant reduction of ARGs detected in the effluent ofprimary clarify, indicating a low removal efficiency of clarify process(p > 0.05). AAO tank was indispensable and basic part in sewage

Fig. 4. Log10 value of absolute abundances of ARGs, 16S rRNA and intI1 in water samples and sludge samples. A & B, ARGs in WWTPs and receiving sea (WE1, WE2,WE3). C, 16S rRNA gene and intI1 in WWTPs and WE (1, 2, 3). Copies/mL, unit of water samples; copies/g, unit of sludge samples.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

5

disposal process to play important role in the removal of a variety ofcontaminant. It has been reported that enhanced hydrolysis could en-able efficient anaerobic treatment of tetracycline production waste-water, with significantly lower generation of ARGs (Yi et al., 2017).However, tetracycline resistance genes could still be detected withabundances of 102-106 copies/mL in the water after conventionalsewage treatment (Auerbach et al., 2007). In this research, the effluentsof AAO process (W1-4 and W2-5) were selected to investigate the re-moval efficiency of AAO process. The absolute abundances of intI1, 16SrRNA gene and ARGs didn’t decrease obviously from the raw influentand the effluent of primary clarify process to the effluent of AAOtreatment (p > 0.05). The result in previous research (Auerbach et al.,2007) was consistent with the result in our research, and provided agood explanation for the high abundances of tet genes (102-106 copies/mL) in the effluent after CAS treatment in our study. Additionally, theabundances of these genes detected in sludge samples were about 3–6orders of magnitude higher than those detected in corresponding was-tewater samples, which demonstrated that sludge was the dominantreservoir for ARGs and microorganism. In the CAS system, the processesof secondary clarify and ultraviolet (UV) disinfection were usually thelast two water treatment plants. The abundances of selected ARGs, intI1and 16S rRNA gene detected in the water after secondary clarify de-creased about 1–2 orders of magnitude. However, the abundances of

these genes detected in the effluent of UV disinfection increased by 1–2orders of magnitude on the contrary, which were at the same level withthose in the discharge of aerobic tank. In total, there was no significantremoval of the detected genes in the CAS system (p > 0.05).

To investigate the removal efficiencies of ARGs, intI1 and 16S rRNAgene in the IMR system, the effluent after microfiltration (W2-7) andRO filtration (W2-8) was chosen to detect the target genes (Fig. 4). Afterthe microfiltration process, the absolute abundances of sul genes (sul1and sul2) and qnrS gene decreased with 2–3 orders of magnitude.Abundances of sul1 and sul2 decreased from 1.86×105 to 2.14× 103

copies/mL and from 1.98×106 to 5.76× 103 copies/mL, respectively,while the abundance of qnrS decreased from 210.33 to 0.46 copies/mL.Meanwhile, abundances of tet (tetB, tetG, tetX), erm (ermF) and qnr(qnrS) genes decreased with 1–2 orders of magnitude. Moreover, nosignificant reduction in the abundances of ermT, qnrA and qnrB acrossmicrofiltration process occurred (p > 0.05). RO membrane permeatetechnology was vital and indispensable unit in the IMR system, whichplayed an important role in desalination and demineralization of waterreuse system (Coutinho de Paula and Amaral, 2017). After RO filtration,the absolute abundances of all detected genes (except qnrB, qnrS and16S rRNA gene) decreased by one order of magnitude. In the effluent ofRO filtration, the abundances of sul, tet, erm, qnr, intI1 and 16S rRNAgene were low by varying from 1.18 to 3.86× 103 copies/mL. The totalcopy number of all selected ARGs was only 4.03×104 copies/mL,which was about 2–3 orders of magnitude lower than that in the rawinfluent of the WWTPs. Interestingly, the absolute abundances of sev-eral ARGs and intI1 in the samples from the influent of the WWTPs (W2-1) to the effluent after microfiltration (W2-7) were higher than those of16S rRNA gene. The rational explanation was that the extracellularDNA (eDNA) in water except the bacteria might survive after thecommon water treatment technologies such as AAO treatment andmicrofiltration (Mao et al., 2014). However, the absolute abundances ofmost of ARGs and intI1 in the effluent after RO filtration (W2-8) werelower than those of 16S rRNA gene, suggesting that the eDNA might beremoved efficiently through the RO filtration.

Removal efficiencies were calculated by log removal value of theraw influent and the effluent after water treatment. As shown inFig. 5A, the removal efficiencies of sul (sul1 and sul2) genes were up to2.1 and 2.5 log removal values after microfiltration process, and up to2.8 and 3.3 log removal values after RO filtration in the IMR system.However, the log removal values of sul1 and sul2 genes were only about0.65 ± 0.04 and 0.06 ± 0.01 in the CAS system, respectively. For tetgenes (tetB, tetG, tetX), the log removal values ranged from 1.6 to 2.5after RO filtration and ranged from 1.0 to 1.6 after microfiltration. Inaddition, the removal efficiencies of erm (ermF, ermT) and qnr (qnrA,qnrB, qnrS) genes were up to 2.3 and 3.8 log removal values after thepermeate of RO membrane, and up to 2.0 and 4.3 log removal valuesafter the permeate of microfiltration membrane. As control, the re-moval efficiencies of erm and qnr genes were poor in the CAS system,lower than 1.4 log removal values. In summary, the removal efficienciesof detected genes after RO filtration were the highest, followed by theremoval efficiencies after microfiltration in IMR system. The removalefficiencies of selected genes after CAS treatment were the lowest.

Daily flux of ARGs, intI1 and 16S rRNA gene through the IMRsystem and the CAS system were illustrated in Fig. 5B. Results indicatedthat the raw influent of the WWTPs was severe breeding ground forARGs, intI1 and 16S rRNA gene by holding the highest daily flux of eachdetected gene. Total daily flux of all detected ARGs was up to(7.22 ± 4.16)×1017 copies/day. Daily flux in the effluent of AAOprocess was the secondary (sum to (7.16 ± 4.83)×1017 copies/day) tothe raw influent, followed by the unit of secondary clarify (SC,(7.19 ± 5.05)×1017 copies/day). Compared with the CAS system,daily flux of the summation of all detected ARGs decreased to(6.09 ± 0.63)×1015 copies/day in microfiltration unit and decreasedsharply to (1.02 ± 1.37)×1014 copies/day in RO filtration process ofthe IMR system, which were 1–3 orders of magnitude lower than that in

Fig. 5. Removal efficiency (A) and daily flux (B) of ARGs, 16S rRNA and intI1 inthe conventional activated sludge system (CAS) and the reclaimed water reusesystem with integrated membrane process (IMR). INF, influent of the WWTPs;AAO, anaerobic-anoxic- aerobic tank; SC, secondary clarified tank; MF, mi-crofiltration process; RO, reverse osmosis filtration process.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

6

the CAS system. This phenomenon indicated that ARGs, intI1 and 16SrRNA gene could be reduced efficiently in the IMR system for waste-water reclamation.

3.3. Distribution of bacterial communities in the IMR system

The diversity index showed a quite different pattern of bacterialcommunity diversity across different samples (Wang et al., 2018).Community richness estimators (Sobs, Chao1 and ACE), communitydiversity estimators (Shannon and Simpson) and coverage were listed inTable S3. The coverage was in the range of 98.97%–99.4%, which re-vealed that the high-throughput sequencing depth was enough to coverthe most of communities as previous reports (Zhao et al., 2018a). In theIMR system, the community richness estimators decreased sharply afterthe treatment of microfiltration (W2-7) and RO filtration (W2-8). Thecommunity diversity estimators also illustrated that the diversity ofwater samples after the process of microfiltration (W2-7) and RO fil-tration (W2-8) were obviously lower than that of other water samples.In contrary, the α-diversity estimators showed that there was no

significant reduction of the richness and diversity in the CAS system.Bacterial communities on phylum level (Fig. S1) and genus level

(Fig. S2) were graphically demonstrated in bar charts. There were ob-vious differences in the bacterial communities of wastewater, sludgeand seawater samples. The relative abundance of Proteobacteria was thehighest on phylum level, followed by Bacteroidetes, Actinobacteria,Acidobacteria, Chloroflexi, Firmicutes, Planctomycetes, Cyanobacteria,Nitrospirae, Synergistetes and so on. In the IMR system, the relativeabundances of Proteobacteria, Bacteroidetes, Actinobacteria andFirmicutes were relatively higher in W2-1 and W2-2. The relativeabundances of Acidobacteria, Planctomycetes and Nitrospirae increased inthe anoxic, aerobic and microfiltration tanks. Interestingly, the relativeabundances of Actinobacteria and Cyanobacteria increased after themicrofiltration and RO filtration processes, while the phyla ofAcidobacteria, Chloroflexi, Firmicutes, Planctomycetes and Nitrospirae de-creased obviously. In total, the relative abundances of Firmicutes andSynergistetes decreased, while the abundances of Actinobacteria andCyanobacteria increased throughout the whole processes of the IMRsystem. It was worthy noted that Cyanobacteria was detected with high

Fig. 6. Canonical correspondence analysis (CCA) analysis based on bacterial community, ARGs, intI1 and 16S rRNA gene (A). Non-metric multi-dimensional scaling(NMDS) analysis (Bray-Curtis distance) based on water quality parameters, bacterial community, ARGs, intI1 and 16S rRNA gene (B).

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

7

relative abundances in receiving seawater samples, which was con-sistent with the bacterial communities of the effluent after micro-filtration process.

The top 87 bacterial genera of all samples were shown in Fig. S2. Inthe surrounding sea of the WWTPs, Pseudoalteromonas accounted for thelargest percentage, followed by genera of Fluviicola, Pelagibacter,Psychrobacter, Actinomarina, Flavobacteriaceae, Alteromonas,Synechococcus, Salinirepens, Aquabacterium and so on. In the raw in-fluent of the IMR system, Comamonas, Macellibacteroides,Cloacibacterium, Rhodococcus, Pseudomonas, Bacteroides, Trichococcus,Geobacter, Desulfobacter and Fulvimarina were the predominant genera,while they disappeared gradually or were detected with low relativeabundances after the treatment of microfiltration and RO filtration.This result indicated that the IMR system contributed to the decrease ofthe bacterial diversity, while the bacterial diversity after the CASsystem did not decrease.

3.4. Correlations of ARGs, bacterial communities and environmental factors

CCA analysis (Luo et al., 2017) among water quality parameters(salinity, pH, ammonium, nitrate, nitrite, phosphate, silicate, TN, TPand TOC), antibiotic resistant genes and bacterial communities weredemonstrated in Fig. 6A. The result indicated that the bacterial com-munities of the seawater samples (WE1, WE2 and WE3) were positivelyaffected by salinity and pH, in accordance with previous research whichdemonstrated that high salinity may be used as a potential method forthe removal of ARGs and bacteria carrying ARGs (Liu et al., 2018). Thewastewater samples collected from the raw influent and primary clarify

unit were positively influenced by TP, TOC, ammonium and phosphate.The result of CCA analysis demonstrated the strong clustering of dif-ferent samples to separate reuse water and the effluent of microfiltra-tion from other water samples.

NMDS analysis between all 22 samples was performed using bray-curtis distances (Zhao et al., 2018b). As shown in Fig. 6B, a strongclustering of the detected ARGs, intI1, 16S rRNA gene and establishedmicrobial communities were constructed according to their abun-dances. The clustering significantly separated seawater samples andsludge samples from the wastewater samples in the WWTPs. Most of all,the reuse water (W2-8) and the effluent of microfiltration (W2-7) weresignificantly separated from other water samples in the WWTPs. Theresult of NMDS analysis was well corresponded with the clustering re-sult of CCA analysis.

3.5. Co-occurrence patterns among ARGs, intI1 and bacterial communities

Co-occurrence patters of ARGs, intI1, 16S rRNA gene and bacterialcommunities were generally investigated by network analysis to revealthe bacterial genera carrying ARGs and intI1 (Guo et al., 2017). Con-nections of significant (p < 0.01) and strong (Spearman’s correlationcoefficient> 0.6) correlations were identified among the top 150 bac-terial genera, ten ARGs, intI1 and 16S rRNA gene (Fig. 7).

ARGs of sul2, tetG, tetX and ermF hold the most frequent correlationswith other genes and bacterial genera. Genera of Rhodoplanes,Mesorhizobium, Pirellula and Pir4_lineage hold the most frequent corre-lations with other genes and genera. Interestingly, 16S rRNA gene waspositively correlated with most of ARGs (sul, tet, ermF and qnrA), intI1

Fig. 7. Network analysis of ARGs, intI1, 16S rRNA and top 150 bacteria genera. Nodes were colored according to bacteria genus and types of genes. The strong(Spearman’s correlation coefficient> 0.6) and significant (p < 0.01) correlation was demonstrated by connections (red, positive correlation; green, negativecorrelation). The size of nodes was proportional to the number of connections. Others, unclassified and norank bacteria in top 150 bacteria genera (For interpretationof the references to colour in this figure legend, the reader is referred to the web version of this article).

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

8

and a few genera of bacteria (Defluviicoccus, Competibacter, Trichococcusand Clostridium, p < 0.01). The intI1 gene was positively correlatedwith all detected ARGs, except ermT and qnrB, which suggested thatintI1 played important roles in the propagation of ARGs. Previous stu-dies also presented that there were not positive and significant corre-lation between intI1 and ermT/qnrB (Chen et al., 2019; Duan et al.,2018; Hu et al., 2019). Meanwhile, intI1 was significantly related to thebacterial genera of Nitrospira, Defluviicoccus, Competibacter, Plancto-myces, Alysiosphaera, Pirellula, Rhodoplanes and Clostridium. Ad-ditionally, there were strong (Spearman’s correlation coefficient>0.77) and significant (p < 0.01) correlations between sulfonamidesantibiotic resistance genes (sul1 and sul2) and tetracyclines antibioticresistance genes (tetB, tetG and tetX).

Seven genes out of all the detected ARGs and intI1 were assigned toClostridium, while six genes were assigned to Defluviicoccus. Moreover,five genes were assigned to Pirellula, Pir4_lineage, Alysiosphaera,Competibacter, Rhodoplanes and Microthrix, while four genes were as-signed to Trichococcus, Tessaracoccus, Ornatilinea, Hyphomicrobium,Mesorhizobium and Nitrospira with positive correlations. All of thesegenera might carry multiple antibiotic resistance genes and be resistantto multiple antibiotics. Furthermore, these bacteria genera, detected inthe effluent of AAO treatment with relatively high abundances, werenot detected in the water samples after microfiltration and RO filtrationor detected with lower relative abundances. For example, Clostridium,which was the potential multi-antibiotic resistant bacteria, accountedfor 0.35% in the raw influent and 0.01% in the effluent of the IMRsystem. Defluviicoccus was also the potential multi-antibiotic resistantbacteria. It was detected with the percentage of 1.26% in the waterafter AAO treatment, and only accounted for 0.01% in the reuse water.In summary, potential multi-antibiotic resistant bacteria could be re-duced effectively by microfiltration and RO filtration processes in theIMR system, while the removal efficiencies of these genera were muchlower in the CAS system.

4. Conclusions

Ten types of ARGs, intI1 and 16S rRNA gene were detected andquantified in the reclaimed water reuse system with integrated mem-brane process in a full-scale WWTPs. Results indicated that ARGs, intI1and 16S rRNA gene could be reduced efficiently in the reclaimed waterreuse system with integrated membrane process, while the removalefficiencies of these genes were lower in the CAS system. CCA andNMDS analysis demonstrated the strong clustering which separated thereuse water from other water samples in the WWTPs. Network analysisrevealed the occurrence of potential multi-antibiotic resistant bacteria.In comparison with conventional wastewater treatment system, thepotential multi-antibiotic resistant bacteria could be removed effec-tively by microfiltration and RO filtration processes in the IMR system.These findings demonstrate that the integrated membrane process is asignificant and efficient approach for the removal of ARGs and multi-antibiotic resistant bacteria.

Declaration of Competing Interest

The authors declare no competing interest.

Acknowledgements

This work was supported by National Natural Science Foundation ofChina (No. 41877131), Taishan Scholar Program of Shandong Province(No. tsqn201812116), One Hundred Talents Program of ChineseAcademy of Sciences (No. Y629041021), and Two-Hundred TalentsPlan of Yantai (No. Y739011021).

Appendix A. Supplementary data

Supplementary material related to this article can be found, in theonline version, at doi:https://doi.org/10.1016/j.jhazmat.2019.121025.

References

Auerbach, E.A., Seyfried, E.E., McMahon, K.D., 2007. Tetracycline resistance genes inactivated sludge wastewater treatment plants. Water Res. 41, 1143–1151.

Berendonk, T.U., Manaia, C.M., Merlin, C., Fatta-Kassinos, D., Cytryn, E., Walsh, F.,Bürgmann, H., Sørum, H., Norström, M., Pons, M.-N., Kreuzinger, N., Huovinen, P.,Stefani, S., Schwartz, T., Kisand, V., Baquero, F., Martinez, J.L., 2015. Tackling an-tibiotic resistance: the environmental framework. Nat. Rev. Microbiol. 13, 310.

Chen, H., Bai, X., Li, Y., Jing, L., Chen, R., Teng, Y., 2019. Characterization and source-tracking of antibiotic resistomes in the sediments of a peri-urban river. Sci. TotalEnviron. 679, 88–96.

Chen, H., Zhang, M., 2013. Occurrence and removal of antibiotic resistance genes inmunicipal wastewater and rural domestic sewage treatment systems in eastern China.Environ. Int. 55, 9–14.

Coutinho de Paula, E., Amaral, M.C.S., 2017. Extending the life-cycle of reverse osmosismembranes: a review. Waste Manag. Res. 35, 456–470.

Duan, M., Gu, J., Wang, X., Li, Y., Zhang, S., Yin, Y., Zhang, R., 2018. Effects of geneticallymodified cotton stalks on antibiotic resistance genes, intI1, and intI2 during pigmanure composting. Ecotoxicol. Environ. Saf. 147, 637–642.

Fiorentino, A., Esteban, B., Garrido-Cardenas, J.A., Kowalska, K., Rizzo, L., Aguera, A.,Pérez, J.A.S., 2019. Effect of solar photo-Fenton process in raceway pond reactors atneutral pH on antibiotic resistance determinants in secondary treated urban waste-water. J. Hazard. Mater. 378, 120737.

Guo, J., Li, J., Chen, H., Bond, P.L., Yuan, Z., 2017. Metagenomic analysis reveals was-tewater treatment plants as hotspots of antibiotic resistance genes and mobile geneticelements. Water Res. 123, 468–478.

Hu, Y., Jiang, L., Zhang, T., Jin, L., Han, Q., Zhang, D., Lin, K., Cui, C., 2018. Occurrenceand removal of sulfonamide antibiotics and antibiotic resistance genes in conven-tional and advanced drinking water treatment processes. J. Hazard. Mater. 360,364–372.

Hu, Q., Zhang, X.X., Jia, S., Huang, K., Tang, J., Shi, P., Ye, L., Ren, H., 2016.Metagenomic insights into ultraviolet disinfection effects on antibiotic resistome inbiologically treated wastewater. Water Res. 101, 309–317.

Hu, Y., Zhang, T., Jiang, L., Luo, Y., Yao, S., Zhang, D., Lin, K., Cui, C., 2019. Occurrenceand reduction of antibiotic resistance genes in conventional and advanced drinkingwater treatment processes. Sci. Total Environ. 669, 777–784.

Lan, L., Kong, X., Sun, H., Li, C., Liu, D., 2019. High removal efficiency of antibioticresistance genes in swine wastewater via nanofiltration and reverse osmosis pro-cesses. J. Environ. Manage. 231, 439–445.

Le, T.H., Ng, C., Tran, N.H., Chen, H., Gin, K.Y.H., 2018. Removal of antibiotic residues,antibiotic resistant bacteria and antibiotic resistance genes in municipal wastewaterby membrane bioreactor systems. Water Res. 145, 498–508.

Li, N., Sheng, G.P., Lu, Y.Z., Zeng, R.J., Yu, H.Q., 2017. Removal of antibiotic resistancegenes from wastewater treatment plant effluent by coagulation. Water Res. 111,204–212.

Li, B., Yang, Y., Ma, L., Ju, F., Guo, F., Tiedje, J.M., Zhang, T., 2015. Metagenomic andnetwork analysis reveal wide distribution and co-occurrence of environmental anti-biotic resistance genes. Int. Soc. Microb. Ecol. J. 9, 2490.

Liu, M., Li, Q., Sun, H., Jia, S., He, X., Li, M., Zhang, X.X., Ye, L., 2018. Impact of salinityon antibiotic resistance genes in wastewater treatment bioreactors. Chem. Eng. J.338, 557–563.

Lu, J., Zhang, Y., Wu, J., Wang, J., Zhang, C., Lin, Y., 2019. Occurrence and spatialdistribution of antibiotic resistance genes in the Bohai Sea and Yellow Sea areas,China. Environ. Pollut. 252, 450–460.

Luo, Y., Mao, D., Rysz, M., Zhou, Q., Zhang, H., Xu, L., Alvarez, P.J.J., 2010. Trends inantibiotic resistance genes occurrence in the Haihe River, China. Environ. Sci.Technol. 44, 7220–7225.

Luo, G., Li, B., Li, L.G., Zhang, T., Angelidaki, I., 2017. Antibiotic resistance genes andcorrelations with microbial community and metal resistance genes in full-scale biogasreactors as revealed by metagenomic analysis. Environ. Sci. Technol. 51, 4069–4080.

Ma, Z., Wu, H., Zhang, K., Xu, X., Wang, C., Zhu, W., Wu, W., 2018. Long-term lowdissolved oxygen accelerates the removal of antibiotics and antibiotic resistancegenes in swine wastewater treatment. Chem. Eng. J. 334, 630–637.

Mao, D., Luo, Y., Mathieu, J., Wang, Q., Feng, L., Mu, Q., Feng, C., Alvarez, P.J.J., 2014.Persistence of extracellular DNA in river sediment facilitates antibiotic resistancegene propagation. Environ. Sci. Technol. 48, 71–78.

Marti, E., Variatza, E., Balcázar, J.L., 2014. Bacteriophages as a reservoir of extended-spectrum β-lactamase and fluoroquinolone resistance genes in the environment. Clin.Microbiol. Infect. 20, O456–459.

Michael, I., Rizzo, L., McArdell, C.S., Manaia, C.M., Merlin, C., Schwartz, T., Dagot, C.,Fatta-Kassinos, D., 2013. Urban wastewater treatment plants as hotspots for the re-lease of antibiotics in the environment: a review. Water Res. 47, 957–995.

Michael-Kordatou, I., Karaolia, P., Fatta-Kassinos, D., 2018. The role of operating para-meters and oxidative damage mechanisms of advanced chemical oxidation processesin the combat against antibiotic-resistant bacteria and resistance genes present inurban wastewater. Water Res. 129, 208–230.

Proia, L., Anzil, A., Borrego, C., Farrè, M., Llorca, M., Sanchis, J., Bogaerts, P., Balcázar,J.L., Servais, P., 2018. Occurrence and persistence of carbapenemases genes in hos-pital and wastewater treatment plants and propagation in the receiving river. J.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

9

Hazard. Mater. 358, 33–43.Pruden, A., Larsson, D.G.J., Amézquita, A., Collignon, P., Brandt Kristian, K., Graham

David, W., Lazorchak James, M., Suzuki, S., Silley, P., Snape Jason, R., Topp, E.,Zhang, T., Zhu, Y.G., 2013. Management options for reducing the release of anti-biotics and antibiotic resistance genes to the environment. Environ. Health Perspect.121, 878–885.

Su, J.Q., An, X.L., Li, B., Chen, Q.L., Gillings, M.R., Chen, H., Zhang, T., Zhu, Y.G., 2017.Metagenomics of urban sewage identifies an extensively shared antibiotic resistomein China. Microbiome 5, 84.

Tahrani, L., Soufi, L., Mehri, I., Najjari, A., Hassan, A., Van Loco, J., Reyns, T., Cherif, A.,Mansour, H.B., 2015. Isolation and characterization of antibiotic-resistant bacteriafrom pharmaceutical industrial wastewaters. Microb. Pathog. 89, 54–61.

Wang, F.H., Qiao, M., Lv, Z.E., Guo, G.X., Jia, Y., Su, Y.H., Zhu, Y.G., 2014. Impact ofreclaimed water irrigation on antibiotic resistance in public parks, Beijing, China.Environ. Pollut. 184, 247–253.

Wang, J.H., Lu, J., Zhang, Y.-X., Wu, J., Zhang, C., Yu, X., Zhang, Z., Liu, H., Wang, W.H.,2018. High-throughput sequencing analysis of the microbial community in coastalintensive mariculture systems. Aquac. Eng. 83, 93–102.

Xu, Y., Xu, J., Mao, D., Luo, Y., 2017. Effect of the selective pressure of sub-lethal level ofheavy metals on the fate and distribution of ARGs in the catchment scale. Environ.Pollut. 220, 900–908.

Yi, Q., Zhang, Y., Gao, Y., Tian, Z., Yang, M., 2017. Anaerobic treatment of antibioticproduction wastewater pretreated with enhanced hydrolysis: simultaneous reductionof COD and ARGs. Water Res. 110, 211–217.

Zhang, Q.Q., Ying, G.G., Pan, C.G., Liu, Y.S., Zhao, J.L., 2015. Comprehensive evaluationof antibiotics emission and fate in the river basins of China: source analysis, multi-media modeling, and linkage to bacterial resistance. Environ. Sci. Technol. 49,6772–6782.

Zhang, T., Hu, Y., Jiang, L., Yao, S., Lin, K., Zhou, Y., Cui, C., 2019. Removal of antibioticresistance genes and control of horizontal transfer risk by UV, chlorination and UV/chlorination treatments of drinking water. Chem. Eng. J. 358, 589–597.

Zhang, Y., Zhuang, Y., Geng, J., Ren, H., Xu, K., Ding, L., 2016. Reduction of antibioticresistance genes in municipal wastewater effluent by advanced oxidation processes.Sci. Total Environ. 550, 184–191.

Zhang, W.J., Xu, X.R., Schwarz, S., Wang, X.M., Dai, L., Zheng, H.J., Liu, S., 2014.Characterization of the IncA/C plasmid pSCEC2 from Escherichia coli of swine originthat harbours the multiresistance gene cfr. J. Antimicrob. Chemother. 69, 385–389.

Zhao, R., Feng, J., Yin, X., Liu, J., Fu, W., Berendonk, T.U., Zhang, T., Li, X., Li, B., 2018a.Antibiotic resistome in landfill leachate from different cities of China deciphered bymetagenomic analysis. Water Res. 134, 126–139.

Zhao, R., Feng, J., Yin, X., Liu, J., Fu, W., Berendonk, T.U., Zhang, T., Li, X., Li, B., 2018b.Antibiotic resistome in landfill leachate from different cities of China deciphered bymetagenomic analysis. Water Res. 134, 126–139.

Zheng, J., Su, C., Zhou, J., Xu, L., Qian, Y., Chen, H., 2017. Effects and mechanisms ofultraviolet, chlorination, and ozone disinfection on antibiotic resistance genes insecondary effluents of municipal wastewater treatment plants. Chem. Eng. J. 317,309–316.

Zhou, L.J., Ying, G.G., Zhao, J.L., Yang, J.F., Wang, L., Yang, B., Liu, S., 2011. Trends inthe occurrence of human and veterinary antibiotics in the sediments of the YellowRiver, Hai River and Liao River in northern China. Environ. Pollut. 159, 1877–1885.

Zhu, Y.G., Zhao, Y., Li, B., Huang, C.L., Zhang, S.Y., Yu, S., Chen, Y.S., Zhang, T., Gillings,M.R., Su, J.Q., 2017. Continental-scale pollution of estuaries with antibiotic re-sistance genes. Nat. Microbiol. 2, 16270.

Zhu, Y.G., Johnson, T.A., Su, J.Q., Qiao, M., Guo, G.X., Stedtfeld, R.D., Hashsham, S.A.,Tiedje, J.M., 2013. Diverse and abundant antibiotic resistance genes in Chinese swinefarms. Proc. Natl. Acad. Sci. U. S. A. 110, 3435–3440.

J. Lu, et al. Journal of Hazardous Materials 382 (2020) 121025

10